Are we in a bubble?

Intro/Disclaimer: Trade at your own risk!

Just like women think “I’m special” and “This time it’ll be different” when it comes to relationships, men think the same way when it comes to stock trading.

Warren Buffett’s advice has remained the same for years: Most people should simply dump their money into an index fund that tracks the market and forget about it rather than actively trade and lose money. I too have recommended that same approach in one of my most-read posts, but given my actions, I think I need to add another oft-repeated piece of advice: “do as I say and not as I do.”

I have been guilty of buying individual stocks instead of an index fund, and my latest rationale for why this is okay is that “I’m providing a service to so many others like me by being the data point showing why it’s a bad idea.” I’ve once again taken something said in jest more seriously than it should have been and published my live returns against the VOO (you can find this in a section below). If you’re thinking of buying individual company stocks, let me once again say, that strategy will almost certainly cause you to lose money in the long run, so may I suggest shipping a tenner my way instead? It’ll save you the time and the stress while generating some profits (okay, fewer losses technically).

Lastly, please assume I don’t know what I am talking about and please just invest in index funds that track the market :). Either way, this post is not financial advice (think of it like financial junk food or financial gossip), and do your due diligence before investing. Anyway, with that huge disclaimer, here are my thoughts on the current market.

Are we in a bubble?

This is the question at the front of everyone’s minds. Many say we are in a bubble because “the prices are too high”. But, how do we know what price is too high? A bubble forms when prices are disconnected from valuations (justified by profits) and pops when liquidity runs out. Stocks rising in price may be justified if their earnings are also rising in proportion. And for some, they are! Companies like NVDA, GOOG, META, MU, SNDK have seen huge increases in both stock prices and profits. Other companies, for example, quantum companies like RGTI, QUBT, etc., seem to be skyrocketing based on… nothing, or not nearly enough. Unfortunately, you can’t really make money off of that since shorting is time-sensitive and as they say, “the market can stay irrational longer than you can stay solvent”. Let’s look at the big factors affecting the market of today.

NVIDIA holds the key

Circular investments

Folks are concerned about the circular investments going on. NVIDIA has stated they’ll invest in OpenAI1, which means - given OpenAI is a customer of NVIDIA - that the money likely flows right back to NVIDIA. This is concerning, but per se does not make a bubble. Per my assessment, it increases the magnitude of the crash we’ll see if either company fails. A counter-argument here is that it is the most rational thing for NVIDIA to do, since the profits currently all flow to them, and if they are able to spread the money around, it boosts the “local AI economy”, allowing these companies to get easy funding to grow. There aren’t many other places that could contribute 500 billion dollars in funding, after all.

NVIDIA’s concentrated customer base

Over 60% of NVIDIA’s revenue comes from just four companies2! Of these four, we can guess three are Google, Amazon and Microsoft, providing access to NVIDIA’s GPUs through their cloud. We don’t know how many of these GPUs are used by the companies themselves and how many are effectively being rented out to others. More GPUs being used by the companies themselves would indicate a higher risk. The fourth company may be OpenAI. If OpenAI fails, that’s ~10% of NVIDIA’s revenue gone without even considering the impacts from others reducing investment in AI.

NVIDIA may not be as profitable

Finally, NVIDIA may not be as profitable as you may think. The money they are owed (accounts receivable) has gone up, similar to what happened to Cisco pre-2000 market crash 3. If their customers (OpenAI, etc.) are unable to pay and go bankrupt, then that could cause NVIDIA stock to crater and the “bubble” to pop.

AI bust?

What might cause the market to crash? A few things, let’s dig into them one at a time.

RoI on AI goes down

Businesses are about making money, and they will buy whatever as long as it can eventually turn a profit. Google and Meta will buy 3 trillion GPUs if they think it’ll turn a profit. Many companies will do the same, even if they don’t see an immediate path to profit, as long as other (ideally big) companies think they’ll see a profit. If Google or Meta (or some other large company) comes out and says they won’t invest in AI anymore as they don’t see the additional investment turning a profit, I think it’d quickly bring things crashing down. NVIDIA stock will crater, followed by everything else given that NVIDIA has provided liquidity by investing in other companies as previously established.

Algorithmic improvements

Taking Deepseek-R1 as an example, better software/algorithms would mean you don’t necessarily need the best chips anymore (I don’t understand this well enough to expand on the exact improvements unfortunately). Better software can reduce the RoI on the latest chips. There’s a lot we don’t know about neural networks yet, so there’s a lot of scope for algorithmic improvements to be found. However, even with a better algorithm, using top-end hardware with the algorithm could provide enough RoI to justify the cost.

Mag7 concentration

Since the Mag7 is the only thing keeping the S&P afloat, if there are similar mistakes being made across this set of seven companies, we’ll effectively see a huge correction or a recession4. For an idea of what these mistakes could be, we’ll turn to Michael Burry (of Big Short fame, known for shorting the market before the 2008 crash).

Burry’s POV

Burry thinks NVIDIA and others are not accounting for GPU depreciation correctly.5 Burry’s view is that many of these GPUs will not be used for as long as these big tech companies claim and newer GPUs will need to be purchased. This means that the costs are higher than currently stated, and so the returns (profits) are actually lower than currently stated. Burry estimates that between 2026 and 2028, these accounting practices could understate depreciation by approximately $176 billion, inflating reported profits across the sector.6

Neoclouds

New companies (like IREN, NBIS) have popped up to build datacenters for AI. There are enough of these companies that the term “neocloud” has been coined to describe them. In contrast, some of the big companies (hyperscalers) - Microsoft in particular7 - have rolled back investments in AI, indicating they see risk and are hedging against a bubble.

Cash flow

There are also concerns about companies’ falling cash reserves due to the amount they are spending on AI. This seems to be somewhat justified when we consider that some of these large companies have sold bonds to finance their AI purchases. This introduces more money into the system and increases the impacts of a crash further8.

Debt

With the rise of sports betting (FanDuel, DraftKings), prediction markets (Polymarket, Kalshi, Robinhood), BNPL (Klarna, Affirm, PayPal, etc.) and margin trading (Robinhood, Interactive Brokers), the amount of debt being used to make purchases seems to be higher than ever before. This lack of regulation is “fine” until the bill comes due. A market crash would mean margin calls and a lot of failed payments which would result in a magnified drop. Just like with subprime loans and the mortgage crisis, this could be the reason the market drops to an extent causing a recession and not just a simple correction.

AI boom!

Well, given all this talk about why we may be in a bubble, what are the reasons for assuming we are not? Simply put, it’s that these fears are overemphasized. Circular investments? Necessary evil (necessary risk), it’s not very dissimilar to how startups use each other’s products. NVIDIA at risk? Well, NVIDIA is still turning a profit, customers are still seeing benefits from AI and money will keep flowing. Countries around the world want NVIDIA GPUs9, so money should keep flowing into the US. NVIDIA has a near monopoly on GPUs and all competitors are based in the US market (GOOG with TPUs, AMD, INTC, etc.), which means that even if NVIDIA is outcompeted, the US economy should still be fine (keep an eye out on China launching their own GPUs given the restrictions). Profitability concerns? Google has stated publicly that they are turning a profit on their AI investments10. The bull will tell you that technology will improve (Blackwell is more power efficient than H100s), which will enable companies leveraging AI to deliver more economic value which will grow the market (total investments into AI). Google has said seven- and eight-year-old TPUs are seeing 100% utilization11, so then, why are we worried about depreciation? The bull case will point to the fact that we are seeing concrete examples of exponential growth. LLMs are now able to perform complex tasks that take humans hours, when earlier it was only able to successfully solve tasks that were a few minutes long12. Could the bull be right?

Conclusion

At the end of the day, the market is effectively an arena for those that want to test their prediction skills. And deciding whether the current market is a “bubble”, I’d say, is an opportunity to do just that. If you think AI will pay off, it’s probably a good time to invest heavily in the market, and specifically in companies involved directly in AI (memory-chip manufacturers, GPU component manufacturers, neoclouds or hyperscalers). If not, it’s a good idea to hold off for now and load up on cash like Burry and Buffett. Of course, some may say it is best to stay invested regardless because the market can stay… eh, you know :).

Bonus: Some tools to help

For those that still wish to try trading, I built a small tool to help estimate what stocks should be priced at: https://www.ashwinmenon.com/finance/stock_pricer/
You can see my returns against VOO here: https://www.ashwinmenon.com/finance/investing/
I’ve also built a BNPL tool, so you can understand the annual rate on these loans if you take them: https://www.ashwinmenon.com/finance/bnpl/
Interesting site to calculate returns with dividends reinvested: https://www.dripcalc.com/compare/intc/voo/
I’ll add further tools under https://www.ashwinmenon.com/finance/, so keep an eye out!

Footnotes

  1. In September 2025, NVIDIA and OpenAI announced a partnership that would include a $100 billion investment over time by NVIDIA into OpenAI. However, NVIDIA has already started walking this back! NVIDIA’s quarterly financial report in November 2025 noted that “there is no assurance that we will enter into definitive agreements with respect to the OpenAI opportunity.” See: “Nvidia says there’s ’no assurance’ of final agreement with OpenAI despite $100 billion pact”, CNBC, November 19, 2025. ↩︎

  2. NVIDIA Corporation SEC filings. See: SEC EDGAR database for NVIDIA Corporation. ↩︎

  3. As of Q3 FY26, NVIDIA’s revenue has increased by 62.5%, growing from $35.082 billion to $57.006 billion. During the same period, accounts receivable grew by 88.7%, rising from $17.693 billion to $33.391 billion. In comparison, Cisco’s revenue grew by 56% and accounts receivable grew by 84% in Fiscal Year 2000. In both cases, accounts receivable has outpaced revenue growth. NVIDIA’s Day Sales Outstanding (DSO) stands at 52+ days (Q3 FY26), whereas Cisco’s DSO was just 40 days in 2000. ↩︎

  4. 54% of S&P 500 gains (trailing twelve months, Q3 2025) are from the Magnificent 7. See: “S&P 500 Index Performance Check — Q3 2025”, FT Portfolios, October 9, 2025. ↩︎

  5. Michael Burry has publicly criticized major tech companies for extending GPU depreciation schedules to overstate earnings. See: “Artificial Intelligencer: The case for selling Nvidia”, Reuters, November 13, 2025. ↩︎

  6. Burry estimates that between 2026 and 2028, tech companies could understate depreciation by approximately $176 billion through extending the useful life of computing assets. See: “Michael Burry warns of $176 billion depreciation understatement by tech giants”, Investing.com. ↩︎

  7. Microsoft has scaled back data center investments globally, signaling an investment re-evaluation. Reports indicate Microsoft halted US and European data center expansions due to oversupply and AI investment skepticism. See: “Microsoft’s huge data center investments face tariffs”, Futurism; “Microsoft pulls back on data centers as AI investments face scrutiny”, Finimize. ↩︎

  8. Major tech companies have issued significant bond offerings to finance AI infrastructure: Alphabet raised $25 billion in bonds (Q4 2025) to fund AI and cloud infrastructure; Meta raised $30 billion in bonds; Amazon issued $15 billion in bonds for AI infrastructure. See: “Alphabet raises $25B in bonds to fund AI and cloud infrastructure”, Wealth Professional; “Funding AI’s trillion-dollar bill — Tech bond issuance soars to bankroll data center spend”, 9fin; “Amazon Issues $15 Billion Bonds for AI Infrastructure”, The Chosun Ilbo, November 18, 2025. ↩︎

  9. Saudi Arabia and NVIDIA announced a major partnership to build AI factories powered by thousands of advanced GPUs. The US has approved billions in NVIDIA GPU exports to the UAE, and the UK has a deal for thousands of Blackwell GPUs. There is also intense demand from China, with NVIDIA lobbying for loosened export controls. See: “Saudi Arabia and NVIDIA to build AI factories to power next wave of intelligence”, NVIDIA Newsroom; “NVIDIA’s CFO says there’s still a ’little geopolitical situation’ before shipping AI GPUs to China”, PC Gamer; “NVIDIA lobbies White House and wins loosened AI GPU export control to China”, Tom’s Hardware. ↩︎

  10. Anat Ashkenazi, SVP and CFO of Alphabet and Google, stated in the 2025 Q3 earnings call: “Yes, and the question related to ROIC and how we look at just overall our business and where do we see early signs that are encouraging. So first, I would say it’s not just early signs, because we’re seeing returns, obviously, in the Cloud business.” See: Alphabet 2025 Q3 Earnings Call, Alphabet Investor Relations. ↩︎

  11. Vahdat, Google’s VP and GM of AI and infrastructure, said that Google currently has seven generations of its TPU hardware in production and that its “seven- and eight-year-old TPUs have 100 percent utilization.” See report ↩︎

  12. Measuring AI performance in terms of the length of tasks AI agents can complete. See METR research ↩︎

Comments

Leave a comment